I'm Shubham Panchal, a passionate programmer specializing in on-device ML and Android development.

• Starting in 2017 as an Android developer, and later picking ML in 2018, the idea of using ML models in Android apps fascinated me from the very beginning.

• I have worked at multiple startups as an Android or ML developer while doing my bachelors in CS (will graduate this year)

• I have several open-source projects, a popular one being SmolChat, an Android app that allows users to run LLMs locally on their device using llama.cpp.

• Other projects include running a RAG pipeline, face recognition and generating sentence embeddings on-device in Android. The projects bear 1170+ stars on GitHub with 250+ forks.

• Challenges in Mobile ML: Integrating a ML model in an Android app has its own challenges, including, memory-constraints, absence of pre/post-processing operations and the ability to reproduce the inference pipeline written in Python.

• Need of Mobile ML: As compute improves on mobile devices, the ability to run these models on-devices helps achieve better hardware utilization, data-privacy and decreased inference latency.

• I also write blogs on Medium and my personal blogpost covering Android, ML, low-level programming.

I will be launching SmolChat here, on ProductHunt in the next few days. Do connect with me on panchalshubhamdev@gmail.com or visit my website shubham0204.github.io to find other social handles.